Author: Michael Neish
Date: 10:15:38 01/28/00
Go up one level in this thread
On January 28, 2000 at 11:49:24, Mark Taylor wrote: >On January 27, 2000 at 22:55:51, Michael Neish wrote: > >Each "game" was a single random "result" - I think what you should have done is >make each *move* a random result. This would have evened out the anomalies by >having a much larger no of events. I disagree completely, since what you're suggesting is that each move is mutually exclusive. Isn't one move influenced by the move that came before it? Furthermore, if Chess games had these characteristics, then probably 99% of all games would be draws. Are they? And besides, if one were to try to model it your way, how many "small errors" need to occur so that one game can be considered won? If Player A makes three small errors, but then Player B makes two, is the game drawn or not? What's the probability of a slightly bigger error occurring? Anyway, Chess doesn't work like this. How many times have you seen two programs playing, both of them show an evaluation close to zero, and then suddenly there's a jump as one of them misses a tactical shot and the other takes advantage of it. After that the game is essentially over. Sometimes the error might not be large enough for a win, and a draw ensues. Other times it may be, and there's a win. How many times have you looked at annotated games and the writer talks of "the losing move"? With grandmasters maybe it just takes one inaccuracy to lose the game. Don't misinderstand me. I'm not saying that there is only one important move that decides the game. I'm saying that your idea of counting each move individually is wrong. As I've said before, get your program to play itself over 20 or 24 games, and see what happens. It's disturbing when one tries to reduce large-scale phenomena to some simple model, I know. But Chess results are rather simple in the scheme of things and I maintain that my original assertions are valid. Cheers, Mike.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.